Agents
Agent workflows that can use your connected tools.
Set up reusable agent workflows with approved org connections, MCP servers, Contexts, and chat history.
Oi Agents gives teams one place to ask for work that spans the tools they have already connected. Instead of copying context between apps, the agent can use the organization's Oi library and connected MCP servers to complete repeatable workflows.
Ask an agent to find related GitHub repositories and draft a launch workflow.
Search Zoho leads, pull supporting Contexts, and summarize what should happen next.
Bundle Oi, GitHub, and Linear into one agent workflow for product triage.
Why it matters
Agents turn connected tools into repeatable operating flows.
Customers should not have to wire together a new local MCP setup every time they want useful work done. Oi can centralize the organization's approved connectors, then make them available to the agent workflow that needs them.
Connector-backed agent workflows
Let teams build workflows that can look up repositories, search CRM data, inspect docs, and use Oi Contexts from the same request.
Org-scoped MCP access
Connections that are successfully set up for the organization can be bundled as MCP servers for the agent query.
Reusable chat history
Keep agent conversations around so teams can return to a workflow, retry errors, and continue from prior context.
How it works
From org connections to a workflow your team can run again.
Step 1
Connect the tools your team already uses
Bring in approved org connections like GitHub, Linear, Zoho, Google Workspace, and Oi itself so an agent can work with the right systems.
Step 2
Describe the workflow once
Turn repeated requests into an agent workflow with the right context, connectors, instructions, and expected outputs.
Step 3
Run it from one chat surface
Ask for the outcome in plain language while Oi bundles the org's available MCP servers behind the scenes.
Step 4
Keep every run tied to the org
Agent chats inherit the organization's connections, guardrails, and Context library instead of relying on one person's local setup.